I have a lot of 5700xt cards from different vendors in the same room and I also have only one PowerColor Red Devil and this card ist the worst card ever. In the same room conditions this card maintain 84C memory temperatur at around 80% cooler for comparison Sapphire Nitro + next to it maintain 76C at 20% cooler. Even the PowerColor Red Dragons are a way better. They hold 80C at around 30% cooler and those with samsung memory even lower 76C at 15-20% cooler. I also noticed that the Gigabyte cards especially Aorus are very bad in memory cooling. I have only 6 of them and they also hold 84C at around 60 - 70% cooler. Does anyone have the same problem and has he managed to solve it? I’m thinking to change the thermal pads but I’m not sure it may go worse. I also don’t know the thikness of thermal pads of Red Devil, Gigabyte Gaming OC or Aorus. Does someone know them ?
Hi, thank you for the information I do have technical expirience ;). I read you comment right after I changed the thermal pads of the card. The ram pads were 2mm but the mostfets were 1mm so I changed all of them with 2mm thermal grizzly Minus 8 pads and now this card is one of the coldest in my set. The difference is huge!
I recommend also not to push you cards so much if you think to keep them for a long period of time, 5-10 yers. I hold my microns at 900 and my samsungs at 885, all with modified timings of course. I beleave you do changed first the timings and then you clock the memory a lot further above 900 to reach 59 MH. This will completely destory your memory in months whatever the temperatures they hold. I do also observed the performance of my cards at 875 885 900 920 even 950 for months now and I can confirm that the best performing cards are those working between 885 - 900. They have the best count of shares and they use less energy which atomatically means less stress for the electonic and longer life.
GPU5 has a samsung memory that produced some bad shares at 800/1300 mV so I decide to test it at 0/0 (850/1350mV) and it works without any bad shares last 30 days so the clock is not the reason fo them and now I’m going to start playing with the voltage
GPU9 for example crashes at 900 885 875 and even 865 so I changed the reaser before 2 days and now it’s stable so I will start playning with it soon and so on
The guy from the post are telling to hold your memory at 1350 mV and even further to 1400mV and this is a joke, 90% of the microns are working just fine at 900 and 800/1300 mV some of them need 1315 or 1325 and that it. They use less energy, produce less heat and expected to life longer. I also do beleve I can set the vddci of my memories below 800 mV, 775mV or even 750mV and they will work just fine. Will check this next days.
I’m new to mining and got an AORUS 5700XT with micron chips, and like you said, it is terrible at memory temps, reading your post I want to understand better the timings you talked about, it is done in the bios mod? I haven’t done any bios related stuff yet.
I’m trying everything to lower memory temps (started mining with it only today), can’t stand it hitting 92ºC (even with 105W power limit, 75% fan, 775 VDDCI and 840 MVDD, the rest at stock), the card is full of thermalpads, but if they are not good will replace them (does it makes me lose warranty? this one still has 3 years). Bios mod also makes me lose warranty?
Really appreciate any help.
I also have 3x 1060 mining while I can’t get new cards at reasonable prices, not the best, but better than nothing hehe.
Use AMDVbFlash to download the bios then use Red Bios Editor to modify the timings and then use AMDVbFlash to flash the modified bios. Search gogole and youtube to learn how to do this. About the warranty If you change the pads I don’t know but if you just modifiy the timings flash the bios and save the original one you can always bring it back in the same way of course if the card is still works and back flash is possible. About the temps I also have some Aorus + some Gaming OC and thay all are bad in memory temps so I’m going to change the thermal pads for all of them next week. Good luck!
After that I’m getting 86~88ºC on memory (better than 94~96ºC before) with PL on 110, 840 memory voltage, 775 on memory controller, 75% fan, not the best world but I guess better than this is hard to get, or maybe only changing the thermalpads. Getting 54 MH/s with this config, didn’t try to change clocks yet.
If you achieve any good improvements on thermalpad changing please let us know. One more thing, what miner are you using? I’m on phoenix right now, is it ok or should i change it?
If anyone else reading this has good experience with this, share the info! hehe
Hi there!
Same card, same problem.
How many thermal pads required for memory and how many for backplate?
What is actual size needed? I can buy 30*30mm piece by piece.
i have 2 red devils, i also got 2mm and applied it. on one card it worked kind of okey-ish. but the temps didnt went down. for the other card they were way too thick. the gpu die didnt touched the cooler surface anymore.
now i will try 1 and 1.5mm. + adding probably a rest to the back
I also do beleve I can set the vddci of my memories below 800 mV, 775mV or even 750mV and they will work just fine. Will check this next days.